Publication number | US7653153 B2 |
Publication type | Grant |
Application number | US 10/440,399 |
Publication date | Jan 26, 2010 |
Filing date | May 19, 2003 |
Priority date | May 17, 2002 |
Fee status | Paid |
Also published as | US20040091058 |
Publication number | 10440399, 440399, US 7653153 B2, US 7653153B2, US-B2-7653153, US7653153 B2, US7653153B2 |
Inventors | Filippo Tosato, Paola Bisaglia |
Original Assignee | Hewlett-Packard Development Company, L.P. |
Export Citation | BiBTeX, EndNote, RefMan |
Patent Citations (11), Non-Patent Citations (14), Referenced by (14), Classifications (19), Legal Events (5) | |
External Links: USPTO, USPTO Assignment, Espacenet | |
The present application is based on, and claims priority from, GB Application Number 0211492.4, filed May 17, 2002, the disclosure of which is hereby incorporated by reference herein in its entirety.
The present invention relates to a method of estimating a reliability measurement for a received bit of a quadrature amplitude modulated (QAM) symbol. The present invention also relates to a receiver and demapper operating in accordance with the method.
A significant amount of effort has been expended in the development of high data rate digital communication. Such communication technologies have brought about wireless local area networks, such as those defined in IEEE802.11a and HIPERLAN/2.
Reducing the computational complexity in the receivers of such a network would be beneficial for implementation of such systems because of the limited resources available in many low cost commercial devices.
In general a first user (which may be human or machine) wishes to transmit a data stream to a second user. The bits in the data stream are firstly convolutionally encoded by an encoder 2 to provide forward error correction. Schemes for convolutional encoding are well known to the person skilled in the art and hence only a brief description is required here. In general convolutional encoding is performed by shifting the data through a chain of delay elements usually implemented as a shift register. In the case of the HIPERLAN/2 and 802.11a standards the convolutional code has a rate ½ and a constraint length of seven.
The chain, as shown in
The arrangement shown in
The encoder 2 produces two data outputs Y_{1 }and Y_{2 }in accordance with the series of connections and taps that are formed within it. Working from the input to the output, the first output is represented by the first, the third, fourth, sixth and seventh taps, or 1011011. The second output Y_{2 }is represented by 1111001. Thus the code generator for this encoder is [133, 171] in octal format. Because the convolutional code has rate ½ the data entering the encoder 2 is a scalar bit stream, while the data leaving the encoder is a stream of binary bit pairs. The data outputs of the encoder 2 are sent to a bit interleaver 10 which spreads the bits so that potential errors caused by channel degradation are seen as independent at the receiver. The interleaved bits are then converted into quadrature-amplitude modulated (QAM) signals by QAM mapper 12 having Gray code labelling.
In other words, bit interleaver 10 and bit mapper 12 cause sequences of 2 m bits to be assigned to one of the M=2^{2m }points in the QAM constellation. The mapping is performed as a Gray coding such that sequences of bits associated with adjacent symbols only differ by one bit.
An exemplary 16-QAM constellation, where m=2, based on the foregoing, is shown in
Bit mapper 12 feeds the complex (real and imaginary parts) signals to orthogonal frequency division multiplexing (OFDM) modulator 14. OFDM is a multi-carrier modulation process that subdivides the frequency spectrum that it uses into a plurality of channels. The process gives good performance over a dispersive transmission path. The data stream is split into N_{sd }lower rate data streams that are transmitted over N_{sd }sub-carriers. The symbol duration increases for the parallel sub-carriers and hence the transmission scheme is more robust in the presence of multi-path interference.
Finally the signal is transmitted by a radio frequency transmitter 16 such that the radio signal propagates via a transmission channel 18 (that in many instances tends to be dispersive and noisy) to a receiver 20. Receiver 20 includes an automatic gain control (AGC) circuit to assure that the receiver average output amplitude is normalised.
Consider the performance of a generic i_{th }sub-carrier used in the OFDM process. The i_{th }channel carries a series of QAM symbols. If we consider just one of these symbols, then
α[i]=α _{I} [i]+jα _{Q} [i] (1)
Where
Each symbol represents a convolutionally encoded bit sequence given by {b_{I,1}; b_{I,2 }. . . ; b_{I,m}; b_{Q,1}; b_{Q,2}; . . . b_{Q,m}}, as shown in
The output of the receiver 20 drives OFDM demodulator 22 whose output is fed to an analog to digital converter (not shown) that drives an equalizer and soft-output QAM demapper 24. Assuming that the cyclic prefix, introduced by the OFDM modulator 14, completely eliminates ISI (Inter OFDM Symbol Interference) and ICI (Inter Channel Interference) and that the channel estimate is error free, the received equalised signal in the i_{th }sub-carrier is given by:
y[i]=r[i]/G _{ch}(i)=a[i]+w[i]/G _{ch}(i) (2)
Where
The de-mapper 24 also receives information from a channel state information (CSI) estimator 26, responsive to an output of demodulator 22. The CSI estimator 26 attempts to deduce, i.e., estimate, the effect of the transmission path of channel 18 on the signal propagating in the channel.
In broad terms channel 18 both scales the QAM signals and rotates them in phase, i.e., the QAM constellation gets rotated. The CSI estimator 26 attempts to estimate the effect of the channel 18 on the symbol. The CSI estimator can work on the assumption that channel 18 is slowly time variant. However, the physical transmission scheme includes “preambles” which contain known sequences of symbols that can be used to estimate the status of channel 18. Thus estimator 26 derives individual phase and modulus estimates for each channel 18 by identifying the preamble.
The output of the demapper 24 drives de-interleaver 28, which in turn drives a convolutional decoder 30, that typically decodes based on the Viterbi algorithm. The Viterbi algorithm of decoder 30 searches the possible code words of the convolutional code and detects the one that is most likely to have generated the received sequence. The search procedure of decoder 30 steps through the code trellis and for each path along the trellis computes a metric which quantifies the discrepancy between the received sequence and the possible coded sequence. If the information associated with bits fed into the decoder is hard (i.e., binary, a sequence of −1's and 1's) the decoder 30 is called a hard decision decoder. Alternatively, if the information is soft, consisting of a hard decision (the sign, i.e., polarity) and a confidence level, or reliability (the magnitude), that represents how much confidence there should be in the hard-decision, the decoder 30 is called a soft-decision decoder. It is well known that soft decision Viterbi decoding can give significant gain over hard decoding at the expense, however, of a greater computational complexity. In order to implement a soft-decision Viterbi decoder, the demapper 24, which precedes the Viterbi decoder 30, needs to deliver soft information associated with the bits.
Soft decision demapping for bit interleaved coded modulation (BICM) signals with BPSK or QPSK modulations is straightforward as the soft bit information, before being weighted by the Channel State Information (CSI), is simply given by the received signals for BPSK and by their quadrature-phase and quadrature components for QPSK. Therefore in the following discussion, consideration is given to the higher modulation formats, for which soft detection requires much more computational effort.
In the literature, two different approaches can be found to calculate the soft information for BICM signals, with multi-level modulations.
One prior art mechanism for bit interleaved coded modulation (BICM) schemes was disclosed by E. Zehavi, “8-PSK Trellis Codes for a Rayleigh Channel” IEEE Trans on Comm, Vol 40, pp 873-884, May 1992. The process starts by calculating sub-optimal bit metrics that are then used inside a Viterbi decoder for path metric computation.
For each bit b_{I,K }and each bit b_{Q,K }(where I and Q represent in-phase and quadrature parts, respectively, and K represents an index of the bit associated with the symbol, where K is an integer in the range 1≦K≦m) the QAM constellation is split into two partitions of complex symbols.
These partitions are
The bit metrics are given by
Finally the metrics are de-interleaved by a de-interleaver 28 and provided as an input to a Viterbi decoder 30.
The Viterbi decoder 30 works according to a well known algorithm which need not be described in detail here. However various web sites give tutorials in Viterbi decoding, such as http://pweb.netcom.com/˜chip.f/viterbi/algrthms2.html.
The convolutional encoder 2 functions as a state machine and the Viterbi decoder 30 is furnished with a state map of the state machine showing which state to state transitions are allowed and which ones are disallowed.
In the second prior art approach, the QAM symbols are first demodulated by a soft output de-mapper and passed to a soft-input Viterbi decoder; see M. Speth et al, “Low Complexity Space-Frequency MLSE for Multi-User COFDM”, IEEE GLOBECOM '99, pp 2395-99, December 1999.
In this approach the process seeks to de-map the received signal into soft bits which have the same sign as that provided by a hard decoder and whose magnitude indicates the reliability of the decision.
The soft bit information assigned to bit b_{I,K }can be shown to be given by the log-likelihood ratio (LLR) of the hard decision on b_{I,K }(see R. Pyndiah et al “Near Optimum Decoding of Product Codes”, IEEE GLOBECOM 94, pp 339-43, November-December, 1994) and can be approximated by
LLR(b_{I,K})Δ|G_{CH}(i)|^{2}·D_{I,K} (5)
We define
as the subset containing the real parts of the complex symbols of subset S_{I,K} ^{(c)}, for c=0.1. It can be shown that equation (4) can be rewritten in a simpler form,
Where the two minima are now taken over real values instead of complex symbols.
If this is, for convenience, explicitly evaluated for the 16 QAM symbols we have
Equivalent expressions hold for the quadrature components with “I” replaced by “Q”.
It has been demonstrated in F. Tosato and P. Bisaglia “Simplified Soft-Output Demapper for Binary Interleaved COFDM with Application to HIPERLAN/2”, IEEE ICC 2002, April-May, 2002, that using the approximate bit metrics M_{C}(b_{I,K}) in equation 3 for path metric calculation inside the Viterbi algorithm is equivalent to demodulating the signals into soft bit values according to equation 6 and then employing a soft Viterbi algorithm for decoding.
The formula for calculating the log likelihood ratio in equation 6 can be further approximated by calculating |D_{I,K}| (or indeed |D_{Q,K}|) as the distance of the received equalised signal y[i] from the nearest partition boundary within the partitioned QAM space and assigning D_{I,K }(or D_{Q,K }as appropriate) the sign “+” or “−” according to the partition in which y[i] falls. The magnitude (i.e., absolute value) is a measure of distance of the received symbol from the decision boundary.
The partitions of
Furthermore by letting d_{I,K }and d_{Q,K }denote half the distance between the partition boundaries B2 and B4 relative to bit b_{I,K }and, b_{Q,K }respectively, then for the 16 and 64 QAM constellations with Gray mapping used in IEEE802.11a and HIPERLAN/2, it can be shown that
In terms of computational complexity the prior art system of decoding the symbols, even with all of the simplifications and approximations invoked, is computationally complex.
To illustrate this complexity, consider the case of a burst transmission. In a burst transmission the channel 18 can be assumed to be time-invariant for the duration of the burst. Thus channel state estimation need only be performed once by the receiver at the beginning of each burst.
If we denote N_{b }the number of bits coded in the data burst and N_{sd }the number of sub-carriers (N_{b}>>N_{SD}) then LLR calculation using the formulae:
LLR(b _{I,K})=|G _{ch}(i)|^{2} ·D _{I,K }
LLR(b _{Q,K})=|G _{ch}(i)|^{2} ·D _{Q,K} (9)
requires one real multiplication per coded bit, plus computation of N_{sd }squared modulus of complex values per physical data burst, which is equivalent to 2N_{sd }real multiplications.
Thus, the approximate LLR calculation requires (N_{b}+2N_{SD}) real multiplications.
According to a first aspect of the present invention a measure of trust of data conveyed by a QAM symbol in a QAM process having at least 16 constellation points is estimated by calculating a linear function of the modulus of a channel state estimation.
It is thus possible to reduce the number of calculations, and in particular multiplications, performed in the soft output de-mapping by using an estimate of likelihood that is derived as a function of |G_{CH}(i)|, rather than |G_{CH}(i)^{2}. Thus, the method significantly reduces the computational task of a receiver responsive to a QAM signal having at least 16 constellations points.
Preferably the estimate of trust for a bit b_{I,K}, for K=1, where K represents a bit index within a complex symbol, is calculated as
{r[i]e^{−jφch(i)}}, where represents the “real” part of a complex number, r[i] is the received symbol in the i_{th }sub-carrier, and e^{−jφch(i) }represents the reciprocal of the phase response of an i_{th }transmission channel over which the symbol was transmitted.Preferably the estimate of trust for a bit b_{Q,K }is calculated as
{r[i]e^{−jφch(i})} where ℑ represents the imaginary part of a complex number, for K=1.Preferably the trust estimate of a bit b_{I,K }is further calculated as −|LLR(b_{I,K−1})|+(|G_{ch}(i)|·d_{I,K}) for K>1, where G_{ch}(i) represents the channel frequency response complex coefficient on an i_{th }channel and d_{I,K }denotes the half distance between partition boundaries in QAM space, for K>1, and the estimate of trust of a bit b_{Q,K }is calculated as −|LLR(b_{Q,K−1})|+(|G_{ch}(i)|·d_{Q,K}) where d_{Q,K }denotes the half distance between partition boundaries in QAM space, for K>1.
Preferably the estimate of trust is approximately a log-likelihood ratio (LLR).
According to a second aspect of the present invention, a measure of trust of data conveyed by a QAM symbol is estimated as a log-likelihood ratio (LLR), where the LLR(b_{I,K}) is calculated as −|LLR(b_{I,K−1})|+(|G_{ch}(i)|·d_{I,K}) for K>1 where G_{ch}(i) represents a channel frequency response complex coefficient on an i_{th }channel and d_{I,K }denotes a half distance between partition boundaries in QAM constellation, and where the LLR(b_{Q,K}) is calculated as −|LLR(b_{Q,K−1})|+(|G_{ch}(i)|·d_{Q,K}) for K>1 where d_{Q,K }denotes the half distance between partition boundaries in QAM constellation, and K represents a bit index within the complex symbol.
The invention also relates to methods and apparatus for deriving the measures of trust, as well as a receiver for deriving an indication of bits in the symbols of the constellations.
The present invention is now further described, by way of example only, with reference to the accompanying figures wherein:
Unlike single carrier systems in which all symbols are affected by the same signal to noise ratio (on average), a multi-carrier OFDM system of the type shown in
However the inventors have realised that using the modulus of the channel frequency response coefficients, instead of the square of the modulus, for calculating “soft bit” information for use by a decoder results in a soft bit approximation causing only a slight performance loss in terms of bit error rate at a given signal strength represented by E_{b}/N_{o}, where E_{b }is the energy per information bit and N_{o }is the power spectral density of the noise.
Approximating the soft bit information in this way results in a computationally efficient process and apparatus employing a one tap equaliser and an LLR calculation subsystem. Thus, the inventors have realised that sub-optimum soft input Viterbi decoding of a binary interleaved coded OFDM signal can be achieved with little additional complexity compared to performing the same operation using hard decoding.
By applying this approximation to the prior art scheme for calculating the LLRs, the following equations are obtained.
Where G _{ch}(i)=|G _{ch}(i)|e ^{jφch(i) }
Similar results hold for other Gray labelling patterns different from that shown in
Thus, compared with the prior art calculations of LLR, it can be seen that inside the LLR expression only the thresholds d_{I,K }and d_{Q,K }are scaled by the coefficients that convey the channel state information.
As a consequence, the following process and apparatus can be adopted for joint OFDM signal equalisation and LLR computation. A block diagram of apparatus for performing the process is illustrated in
Fourier transform block 50 converts the incoming OFDM signal from receiver 20 into individual data channels 1 to N_{sd}. Channels 1 to N_{sd }respectively derive signals representing the components in frequency channels 1 . . . N_{sd }of the output of receiver 20. The output of block 50 for channel i represents an approximate value of a symbol of a particular QAM constellation that is applied to a processor arrangement including channel frequency response estimator 52, multiplier 54(i), soft bit calculator 56(i) and scalar calculator 58(i).
The processor arrangement initially equalises/corrects the phases of outputs of blocks 50. This is done by sending each one of the channels 1 to N_{sd }derived by block 50 to an input of a channel frequency response estimator 52 which estimates the phase shift introduced by a link between transmitter 16 and receiver 20 of each one of the channels and thereby produces a phase correction signal e^{−jφch(i) }for each channel i, where i is an integer in the range 1≦i≦N_{sd}. Response estimator 52 performs a channel state estimate once per physical burst, at the beginning thereof.
The phase corrections that estimator 52 derives are applied to each of the channels via respective multipliers 54(1) to 54(N_{sd}). The resulting phase equalised channel signals at the outputs of multipliers 54(1) to 54(N_{sd}) are then supplied to first inputs of respective soft bit calculators 56(1) to 56(N_{sd}) (illustrated in
The channel frequency response estimator 52 also estimates the modulus of the signal strength in each one of the channels 1 to N_{sd}. Estimator 52 supplies the modulus estimate for each of channels 1 to N_{sd }(i.e., |G_{ch}(i)|, where i is selectively 1 . . . N_{sd}) to an associated threshold calculator 58(1) to 58(N_{sd}) for each individual channel. Each of calculators 58(1) to 58(N_{sd}) calculates the real and imaginary threshold values |G_{ch}(i)|·d_{I,K }and |G_{ch}(i)|·d_{Q,K}, where i represents the channel number. Each of calculators 58(1) to 58(N_{sd}) scales its input from estimator 52 by the predetermined stored value or values d_{k }to derive its associated threshold value or values.
For a square QAM constellation having 16 symbols, each of which represents four bits, as illustrated in
Each of soft bit calculators 56(1) to 56(N_{sd}) derives the 2 m bit values respectively associated with each of the symbols 1 to N_{sd }that block 50 derives. For the 16 QAM constellation, wherein m=2, as discussed and illustrated in connection with
The specific circuitry or computer processes included in each of the soft bit calculators 56(1) to 56(N_{sd}) is shown in
For the 16 QAM constellation arrangement shown in
The arrangement shown in
Comparing the complexity of the apparatus of
However, these multiplications are not actual multiplications because for square QAM constellations (e.g. a 16 or 64 QAM), d_{K }are powers of two; hence in fact only bit-wise shifts are needed for threshold scaling.
However, in order to calculate N_{sd }moduli of the complex values per physical burst, it is necessary to calculate the channel state information values. This requires 2N_{sd }real multiplications and N_{sd }square roots to be formed.
Comparing the computational workload:
Arrangement of FIG. 5=2N_{sd }real multiplications and N_{sd }square roots.
Prior art=2 N_{sd }real multiplications and N_{b }real multiplications.
The computational advantage of the arrangement of
The performance of the arrangement of
It should be noted that, for simplicity, the specific embodiments of
Cited Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|
US5134635 * | Jul 30, 1990 | Jul 28, 1992 | Motorola, Inc. | Convolutional decoder using soft-decision decoding with channel state information |
US6078626 | Sep 24, 1997 | Jun 20, 2000 | Ericsson Inc. | Methods and systems for communicating information using separable modulation constellations |
US6115435 * | Oct 30, 1997 | Sep 5, 2000 | Advanced Digital Television Broadcasting Laboratory | Soft decision method and receiver |
US6243423 * | Jun 24, 1998 | Jun 5, 2001 | Sony Corporation | Receiver, transmitter-receiver, and communication method |
US6377607 * | May 13, 1999 | Apr 23, 2002 | Qualcomm Incorporated | System and method for performing accurate demodulation of turbo-encoded signals via pilot assisted coherent demodulation |
US6594318 * | Mar 8, 2000 | Jul 15, 2003 | Qualcomm Incorporated | Method and apparatus for computing soft decision input metrics to a turbo decoder |
US6977972 * | Jul 12, 2000 | Dec 20, 2005 | Sharp Laboratories Of America, Inc. | Method of hybrid soft/hard decision demodulation of signals with multilevel modulation |
US6980602 * | Jan 31, 2001 | Dec 27, 2005 | Comsys Communication & Signal Processing Ltd. | Normalization of equalizer soft output for channels with varying noise power |
US7076000 * | Jan 18, 2001 | Jul 11, 2006 | Motorola, Inc. | Soft-decision metric generation for higher order modulation |
US20020037057 | Oct 15, 2001 | Mar 28, 2002 | Kroeger Brian William | Adaptive weighting method for orthogonal frequency division multiplexed soft symbols using channel state information estimates |
US20030185319 * | Mar 30, 2002 | Oct 2, 2003 | Kolze Thomas J. | Modified branch metrics for processing soft decisions to account for phase noise impact on cluster variance |
Reference | ||
---|---|---|
1 | "Average Power Analysis," http://www.ece.utexas.edu/~bevans/courses/realtime/lectures/15-QAM/lecture15/node5.html, May 17, 2003. | |
2 | "Performance Analysis of PAM," http://www.ece.utexas.edu/~bevans/courses/realtime/lectures/15-QAM/lecture15/node3.html, May 17, 2003. | |
3 | "Performance Analysis of QAM," http://www.ece.utexas.edu/~bevans/courses/realtime/lectures/15-QAM/lecture15/node4.html, May 17, 2003. | |
4 | "Phase Shift by 90 degrees," http://www.ece.utexas.edu/~bevans/courses/realtime/lectures/15-QAM/lecture15/node2.html, May 17, 2003. | |
5 | "Average Power Analysis," http://www.ece.utexas.edu/˜bevans/courses/realtime/lectures/15—QAM/lecture15/node5.html, May 17, 2003. | |
6 | "Performance Analysis of PAM," http://www.ece.utexas.edu/˜bevans/courses/realtime/lectures/15—QAM/lecture15/node3.html, May 17, 2003. | |
7 | "Performance Analysis of QAM," http://www.ece.utexas.edu/˜bevans/courses/realtime/lectures/15—QAM/lecture15/node4.html, May 17, 2003. | |
8 | "Phase Shift by 90 degrees," http://www.ece.utexas.edu/˜bevans/courses/realtime/lectures/15—QAM/lecture15/node2.html, May 17, 2003. | |
9 | Ryndiah et al., "Near Optimum Decoding of Product Codes," pp. 339-343, 1994, IEEE. | |
10 | Speth et al., "Low Complexity Space-Frequency MLSE for Multi-User COFDM," Global Telecommunications Conference-Globecom '99, pp. 2395-2399, 1999, IEEE. | |
11 | Speth et al., "Low Complexity Space-Frequency MLSE for Multi-User COFDM," Global Telecommunications Conference—Globecom '99, pp. 2395-2399, 1999, IEEE. | |
12 | Ström, "Notes on Quadrature Amplitude Modulation," Document No. ESS140/ext:10, rev.: A, Oct. 22, 2003, 12 pp. | |
13 | Tosato et al., "Simplified Soft-Output Demapper for Binary Interleaved COFDM with Application to HIPERLAN/2," pp. 664-668, IEEE, 2002. | |
14 | Zehavi, "8-PSK Trellis Codes for a Rayleigh Channel," IEEE Transactions on Communications, pp. 873-894, May 1992, vol. 40, No. 5. |
Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|
US8139612 | Apr 4, 2006 | Mar 20, 2012 | Qualcomm Incorporated | Methods and apparatus for dynamic packet mapping |
US8213548 | Apr 4, 2006 | Jul 3, 2012 | Qualcomm Incorporated | Methods and apparatus for dynamic packet reordering |
US8413031 * | Dec 16, 2008 | Apr 2, 2013 | Lsi Corporation | Methods, apparatus, and systems for updating loglikelihood ratio information in an nT implementation of a Viterbi decoder |
US8503552 * | Apr 15, 2008 | Aug 6, 2013 | Fundacio Centre Tecnologic De Telecomunicacions De Catalunya | Method and digital communication device for receiving data using QAM symbols |
US8654910 | Aug 9, 2012 | Feb 18, 2014 | Marvell World Trade Ltd. | Co-channel interference cancellation with multiple receive antennas for BICM |
US8971815 | Apr 21, 2014 | Mar 3, 2015 | Marvell International Ltd. | Selecting modulation and coding scheme in the presence of interference |
US9143785 | Oct 25, 2012 | Sep 22, 2015 | Allen LeRoy Limberg | COFDM broadcast systems employing turbo coding |
US9144034 | Sep 13, 2013 | Sep 22, 2015 | Alcatel Lucent | Methods and apparatus for reducing interference effect on data transmitted over a communication channel |
US9154238 | Mar 29, 2013 | Oct 6, 2015 | Alcatel Lucent | Methods and apparatus for reducing interference effect on data transmitted over a communication channel |
US20070230490 * | Apr 4, 2006 | Oct 4, 2007 | Jinxia Bai | Methods and apparatus for dynamic packet mapping |
US20070230632 * | Apr 4, 2006 | Oct 4, 2007 | Jinxia Bai | Methods and apparatus for dynamic packet reordering |
US20100150280 * | Dec 16, 2008 | Jun 17, 2010 | Gutcher Brian K | METHODS, APPARATUS, AND SYSTEMS FOR UPDATING LOGLIKELIHOOD RATIO INFORMATION IN AN nT IMPLEMENTATION OF A VITERBI DECODER |
US20110038406 * | Apr 15, 2008 | Feb 17, 2011 | Stephan Pfletschinger | Method and digital communication device for receiving data using qam symbols |
US20140294126 * | Mar 29, 2013 | Oct 2, 2014 | Alcatel-Lucent Usa Inc. | Methods and apparatus for reducing interference effect on data transmitted over a communication channel |
U.S. Classification | 375/341, 375/346 |
International Classification | H04L25/02, H04L27/06, H03M13/45, H04L27/26, H04L25/03, H04L25/06 |
Cooperative Classification | H04L25/067, H04L2025/0342, H04L25/022, H04L25/03318, H04L2025/03414, H04L27/2647, H04L27/38 |
European Classification | H04L25/03B7K1, H04L25/06C, H04L27/38, H04L25/02C5 |
Date | Code | Event | Description |
---|---|---|---|
Nov 12, 2009 | AS | Assignment | Owner name: HEWLETT-PACKARD DEVELOPMENT COMPANY, L.P., TEXAS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:TOSATO, FILIPPO;BISAGLIA, PAOLA;REEL/FRAME:023511/0616;SIGNING DATES FROM 20091001 TO 20091103 |
Sep 6, 2013 | REMI | Maintenance fee reminder mailed | |
Sep 16, 2013 | FPAY | Fee payment | Year of fee payment: 4 |
Sep 16, 2013 | SULP | Surcharge for late payment | |
Nov 9, 2015 | AS | Assignment | Owner name: HEWLETT PACKARD ENTERPRISE DEVELOPMENT LP, TEXAS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HEWLETT-PACKARD DEVELOPMENT COMPANY, L.P.;REEL/FRAME:037079/0001 Effective date: 20151027 |